Wan 2.5 API

Turn short prompts or reference images into ready-to-post videos with synced audio using Wan 2.5 API on Evolink AI.

Model Type :

Prompt

Estimated Cost

6 Credits

Sample Result

No sample available

Prompt

0 (suggested: 2,000)

Input Images *

Upload 1 image (max 10MB)

Click to upload or drag and drop

Supported formats: JPEG, JPG, PNG, BMP, WEBP
Maximum file size: 10MB; Maximum files: 1

Duration

Quality

Click Generate to see preview

History

Max 20 items

0 running · 0 completed

Your generation history will appear here

Wan 2.5 API for AI video with sound

Generate short HD videos with native audio, lip-sync, and social-ready framing from simple text or image inputs, all via a clean Evolink AI API.

Social media showcase of AI video generation product feature 1

What is Wan 2.5 API on Evolink AI?

Text-to-video with audio

Wan 2.5 API lets you send a short text prompt and receive a cinematic video clip complete with auto-generated sound, voice, or ambience so your content is ready for TikTok, Reels, and Shorts without extra editing. Instead of stitching tools together, you get visuals, pacing, and audio in one pass, which makes it easy to test hooks, concepts, and ad angles at scale across different social media accounts.

Use Wan 2.5 text to video

Vertical feed showcase of AI video generation product feature 2

Image-to-video for product stories

With Wan 2.5 API you can upload a single key visual, such as a product shot or character design, and turn it into a short, dynamic clip that still feels consistent with the original image. This works well for turning static catalog images into scroll-stopping ads, motion posters, or story snippets where the camera moves, the light changes, and audio reinforces your brand message in a few seconds.

Animate images with Wan 2.5

Product motion showcase of AI video generation product feature 3

Built for social-first creators

Wan 2.5 API via Evolink AI is designed around social media creators and indie SaaS builders who care about speed, volume, and consistency more than academic benchmarks. It focuses on short 3–10 second clips, vertical and square formats, and audio that actually feels aligned with what shows on screen so your videos can drop straight into content calendars, UGC templates, or automated posting systems without hand-fixing every render.

Scale social content with Wan 2.5

Creator workflow showcase of AI video generation product feature 4

Why choose Wan 2.5 API via Evolink AI?

Wan 2.5 API combines Alibaba’s audio-visual model with Evolink AI’s simple routing so you focus on ideas, not cloud configs or complex infrastructure.

Audio and video in one pass

Most AI video tools still make you juggle separate models for visuals and sound, which introduces friction, file management, and sync issues. Wan 2.5 API generates video and audio together, including lip-sync and ambient sound, so what you get already feels like a finished short-form clip. For creators and SaaS products that live or die by speed, that one-step workflow removes a lot of invisible overhead.

Simple Wan 2.5 integration

Wan 2.5 itself is an Alibaba Tongyi Wanxiang model, but Evolink AI wraps it into a straightforward Wan 2.5 API so you do not need to deal with region settings, separate console projects, or complex billing dashboards. You call a single endpoint with clear parameters and Evolink AI handles routing to the underlying Wan 2.5 model, making it much easier for developers and creators to plug video generation into their products or content workflows.

Optimized for short-form experimentation

Wan 2.5 API focuses on short, punchy clips in HD so you can rapidly iterate creative ideas instead of waiting for long renders. For marketers, agencies, and growth teams, this aligns perfectly with constant testing across audiences, geos, and hooks. You can generate many small, targeted variations, see what performs, and reinvest in the concepts that actually move metrics.

How to use Wan 2.5 API

Connect Wan 2.5 API through Evolink AI and move from prompt to published video in a few simple steps.

Connect your Evolink AI account

Sign up or log into Evolink AI, create an API key, and enable Wan 2.5 API access so your app can securely call the video generation endpoints without touching Alibaba Cloud directly.

Send prompts, images, and basic settings

Choose text-to-video or image-to-video, write a clear prompt, upload an optional reference image, set duration and aspect ratio, then send a simple JSON request to the Wan 2.5 API route.

Receive, review, and publish your clips

Fetch the generated Wan 2.5 video URL, preview audio and visuals, then plug it into your editor, scheduler, or SaaS interface for immediate download, posting, or further automation.

View Wan 2.5 API docs

Key Wan 2.5 API features

Wan 2.5 API on Evolink AI focuses on real-world social and marketing use cases rather than lab demos, so every feature maps to a clear creator benefit.

Audio-visual

Native audio and lip-sync

Wan 2.5 API can render video with audio by default, including voices, effects, or music, which means your team no longer needs a separate soundtrack pipeline just to make clips feel alive.

Short-form

Short HD clips for social feeds

The model is tuned for 3–10 second HD videos so you hit the sweet spot for TikTok, Reels, and ad placements without wasting budget or time on overly long renders that nobody watches.

Flexible input

Text or image as flexible input

You can start from a simple text script or reuse an existing image as your base, which lets you adapt Wan 2.5 API to ideation, product showcases, and creator tools inside the same stack.

Global-ready

Multi-language prompt and audio support

Wan 2.5 API is comfortable with Chinese and English prompts and can keep audio aligned, which is especially useful when your audience spans multiple regions and languages online.

Cinematic feel

Consistent motion and control

The model offers smoother motion and better camera dynamics than older Wan versions, so videos feel more cinematic and less like janky demos, even when you move fast on campaigns.

Automation

Built-in for automation and SaaS

Because Wan 2.5 API runs through Evolink AI, you can plug it into cron jobs, no-code tools, or full SaaS backends to auto-generate video assets based on schedules, feeds, or prompts.

Wan 2.5 API vs other AI video models

Compare Wan 2.5 API with leading AI video backbones on cost, duration, and ideal use cases so you can choose the right model for each project.

Model	Duration	Resolution	Price	Strength
Wan 2.5 API	3–10 second clips focused on short-form hooks and social stories	Up to 1080p HD with lower tiers at 480p and 720p for budget control	Around $0.05 per second for HD video generation in many pay-per-use setups	Balanced quality, cost, and speed with native audio and lip-sync for social-first workflows
Kling 2.6	5–10 second clips, with options for longer high-motion shots	Up to 1080p with strong motion realism and physics for complex scenes	Commonly around $0.07–$0.14 per second depending on resolution and priority tier	Very strong motion quality and physics, good for realistic avatar videos and dynamic product shots
Seedance 1.5 Pro	4–12 second audio-video clips with flexible dialogue settings	Up to 1080p with tightly synchronized audio and video	Often positioned near $0.05 per second for 720p audio-video generation in competitive offerings	Joint audio-video model with precise lip-sync and dialogue control, great for talking heads and explainers
Sora 2	10–20 second cinematic clips suitable for hero assets	720p to 4K with high-end cinematic quality and detailed motion	Typical guidance puts standard Sora 2 around $0.10 per second for 720p, with higher rates for 1080p and 4K	Top-tier realism and storytelling power for flagship campaigns and premium branded content
Veo 3	Short to mid-length clips tuned for cinematic storytelling	High-resolution output up to 4K depending on provider and plan	Frequently listed close to $0.40 per second for higher-end video generations in external pricing tables	High-end cinematic aesthetic suitable for trailers, launch videos, and professional creative studios

Wan 2.5 API FAQ

Everything you need to know about the product and billing.

Wan 2.5 is an AI video model developed by Alibaba’s Tongyi Wanxiang team that turns text prompts or single images into short HD videos with native audio and lip-sync. Wan 2.5 API via Evolink AI is an integration layer that exposes this model through one user-friendly API, so you can call Wan 2.5 from your apps, SaaS products, or creator tools without touching Alibaba Cloud directly. It is designed for developers, marketers, and social media creators who want reliable audio-video generation as a service rather than managing low-level infrastructure.

Wan 2.5 API is best at short, story-like clips that fit social platforms and performance ads. You can use it to produce product explainers, scroll-stopping intros, motion posters, and quick story beats that pair visuals with sound. Many teams use Wan 2.5 as a way to test hooks and concepts before investing in full productions, generating multiple variants of the same idea and letting real engagement data decide which direction to scale.

Wan 2.5 API delivers a strong balance of audio-visual quality, cost, and speed for 3–10 second videos, while newer models such as Wan 2.6 push further into multi-shot storytelling and advanced narrative control. For many day-to-day social and UGC tasks, Wan 2.5 already provides enough realism and motion to feel native in feeds, at a friendlier price point than heavy cinematic models. That makes it a good fit as a workhorse model in your stack, especially when combined with higher-end options for premium shots.

Wan 2.5 API supports both text-to-video and image-to-video modes, which you can choose through a parameter in the Evolink AI endpoint. In text mode, you provide a prompt describing scene, style, and pacing, and the model generates a fresh clip from scratch with appropriate audio. In image mode, you upload a still frame, key visual, or character and let the video animate around it, preserving identity while adding camera moves, transitions, and sound.

Typical Wan 2.5 API clips run in the 3–10 second range, which is ideal for hooks, transitions, and bite-sized stories on social. Resolution options commonly span from 480p and 720p up to full 1080p, depending on your chosen settings and plan. This keeps render times and costs manageable while still delivering visuals that look clean on modern phone screens and ad placements in vertical or square formats.

Wan 2.5 API is typically billed on a per-second basis through usage-based credits, with pricing aligned to roughly mid-market levels for HD video generation. This means shorter clips and lower resolutions use fewer credits, while longer or higher-resolution clips consume more, giving you control over budget and quality. The result is a pay-for-what-you-generate model that works well for experiments, campaigns, and SaaS workloads with changing demand.

Wan 2.5 API via Evolink AI is designed with commercial use in mind, from performance ads to branded content and embedded SaaS features. As with any AI service, you should still review the latest terms from both Evolink AI and the underlying provider, check usage guidelines for your region and vertical, and put a simple internal review step in place before launching client campaigns or high-visibility assets.

You can call Wan 2.5 API from backend services, serverless functions, low-code tools, or even no-code automation platforms that support HTTP requests. A common pattern is to trigger video generation when new content is created, when a user submits a brief, or on a schedule for recurring campaigns. Evolink AI’s documentation includes example requests, response objects, and error patterns so your engineering time goes into product logic instead of debugging raw infrastructure.

API Reference

Select endpoint

Authentication

All APIs require Bearer Token authentication.

Header

Authorization: 
Bearer YOUR_API_KEY

Get API Key

POST

/v1/videos/generations

Create Video

Wan 2.5 Video Image to Video (wan2.5-image-to-video) model supports image-to-video generation mode.

Asynchronous processing mode, use the returned task ID to .

Generated video links are valid for 24 hours, please save them promptly.

Request Parameters

modelstringRequiredDefault: wan2.5-image-to-video

Video generation model name.

Examplewan2.5-image-to-video

promptstringRequired

Prompt describing what kind of video to generate from the input image.

Notes

Limited to 2000 tokens

ExampleA cat playing piano

durationintegerOptional

Duration of the generated video (seconds).

Value	Description
5	5 seconds
10	10 seconds

Notes

Pre-charged based on duration, actual charge based on generated video duration

Example5

qualitystringOptionalDefault: 720p

Video quality.

Value	Description
480p	Lower quality, lower price
720p	Standard quality (default)
1080p	High quality, higher price

Example720p

image_urlsarrayRequired

Reference image URL list for first-frame image-to-video feature.

Notes

1 image required for image-to-video generation
Max size: 10MB per image
Formats: .jpeg, .jpg, .png (no transparent), .bmp, .webp
Resolution: width and height range [360, 2000] pixels
URLs must be directly viewable by the server

Examplehttps://example.com/image1.png

prompt_extendbooleanOptionalDefault: true

Whether to enable intelligent prompt rewriting.

Notes

When enabled, a large language model will optimize the prompt
Effective for prompts that lack detail or are too simple

Exampletrue

callback_urlstringOptional

HTTPS callback address after task completion.

Notes

Triggered on completion, failure, or cancellation
Sent after billing confirmation
HTTPS only, no internal IPs
Max length: 2048 chars
Timeout: 10s, Max 3 retries

Examplehttps://your-domain.com/webhooks/video-task-completed

Request Example

{
  "model": "wan2.5-image-to-video",
  "prompt": "A cat playing piano",
  "duration": 5,
  "quality": "720p",
  "image_urls": [
    "https://example.com/image1.png"
  ]
}

Response Example

{
  "created": 1757169743,
  "id": "task-unified-1757169743-7cvnl5zw",
  "model": "wan2.5-image-to-video",
  "object": "video.generation.task",
  "progress": 0,
  "status": "pending",
  "task_info": {
    "can_cancel": true,
    "estimated_time": 120,
    "video_duration": 8
  },
  "type": "video",
  "usage": {
    "billing_rule": "per_call",
    "credits_reserved": 7,
    "user_group": "default"
  }
}